- Email: [email protected]
The interaction of rivers and urban form in mitigating the Urban Heat Island effect: A UK case study
Building and Environment 58 (2012) 14e22
Contents lists available at SciVerse ScienceDirect
Building and Environment journal homepage: www.elsevier.com/locate/buildenv
The interaction of rivers and urban form in mitigating the Urban Heat Island effect: A UK case study E.A. Hathway a, *, S. Sharples b a b
Department of Civil and Structural Engineering, University of Sheffield, Sir Frederick Mappin Building, Mappin Street, Sheffield, South Yorkshire S1 3JD, UK School of Architecture, University of Liverpool, Liverpool, UK
a r t i c l e i n f o
a b s t r a c t
Article history: Received 16 March 2012 Received in revised form 19 June 2012 Accepted 20 June 2012
The Urban Heat Island (UHI) effect already produces elevated temperatures in city centres therefore urban design has a key role to pay in reducing the UHI to create safe and pleasant places in which to live and work. Increased surface porosity and bodies of surface water have a role to play in increasing potential cooling through evaporation. Urban rivers may, therefore, have a place in reducing the UHI. This paper investigates the effectiveness that small urban rivers may have in reducing the UHI effect and also examines the role that the urban form on the banks of a river can play in propagating or reducing this potential cooling. The results from a field survey during spring and summer are presented for a river in Sheffield, UK. The level of cooling is related to the ambient air temperature, increasing at higher temperatures. However, there are also seasonal dependencies and relationships linked to the river water temperature, incident solar radiation, wind speed and relative humidity. A mean level of daytime cooling of over 1.5 C was found above the river in spring, but this was reduced in summer when the river water temperature was warmer. The urban form on the river bank influenced the levels of cooling felt away from the river bank. Ó 2012 Elsevier Ltd. All rights reserved.
Keywords: Urban Heat Island Rivers Microclimate Blue infrastructure
1. Introduction It is becoming increasingly important to provide resilience to the urban infrastructure in order to withstand extreme events brought about by a changing climate. Flooding, high wind speeds and increased air temperatures all present different risks to cities. Increases in mean air temperatures and more frequent heat waves means the Urban Heat Island (UHI) is becoming an increasingly significant problem in the UK. An overview of the impacts of the UHI in the UK and methods for mitigation and climate change resilience has been reviewed by Smith and Levermore [1]. Globally, there is a significant body of research attempting to understand and quantify local microclimates, often with the aim of providing design guidance to generate better quality urban spaces. Of particular interest is the provision of vegetation and green spaces; both large and small parks have been found to provide cooling, with the effects propagating to a distance approximately half the park width away, dependent upon the local street layout [2e5]. The provision of large canopy street trees can offset the heat input from vehicles [6] and the choice of built environment surfaces with high solar
* Corresponding author. Tel.: þ44 (0) 114 2225702; fax: þ44 (0) 114 2225700. E-mail address: a.hathway@sheffield.ac.uk (E.A. Hathway). 0360-1323/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.buildenv.2012.06.013
reflectances (albedo) can result in lower levels of solar absorption and so cooler surfaces and adjacent air temperatures [7]. Alongside the provision of shading and cooling via evapotranspiration from vegetation, green spaces will also usually improve the surface porosity, thereby increasing the available capacity for water storage and so water availability for evaporative cooling. The reintroduction of water through the deliberate incorporation of porous surfaces, e.g. porous paving, or the presence of water bodies, such as ponds or rivers, has the potential to reduce the UHI by returning the surface moisture availability to values similar to rural areas. The process of evaporation has been studied, and resulting cooling for various cities demonstrated with models validated for a range of locations [8]. However, there is limited published research on the microclimate effects in the immediate locality of river corridors, with current published data limited to tropical climates. The process of daylighting a large stretch of watercourse in Seoul, Korea, provided the opportunity of a before and after study demonstrating cooling of the urban microclimate [9]. Another study in Nanjing, China [10] of the urban climate highlighted the cooling effects in the locality of a lake, river and sea. The cooling effect of the Ota River, Hiroshima, Japan was found to reach up to 5 C directly above the river and propagated nearly 100 m from the river banks [11]. These studies demonstrate the potential cooling from rivers in hot climates. Further analysis of the
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
influence of local urban form on the propagation of this cooling would be beneficial for providing guidance on river corridor regeneration for an improved microclimate. Although there are limited published studies on the cooling effects rivers provide in Europe, the heat flux between the atmosphere and rivers has been studied extensively in relation to the thermal properties of the river water. This is mainly due to the resulting ecological implications such as fish spawning [12,13]. The heat budget of a river is related to the sensible, radiative and evaporative heat fluxes between it and the surroundings. The process resulting in the greatest effect is dependent upon the local environment. The direction of heat flux varies during the diurnal cycle, for instance absorbing solar radiation during the day, and radiating long wave radiation at night. However, it seems the radiative heat attenuated and absorbed by the river during the day is greater than that released to the environment during the evening and night, at least for rural areas [12]. In comparison to static absorption and release of radiation for most urban materials, the heat absorbed by the water and suspended particles will be carried downstream at a rate dependent on the river dynamics; heat absorbed will then be released at a different location downstream. The albedo of river water also varies with the angle of incidence of the solar radiation and the quantity of suspended particles. Therefore, intense rain events, resulting in large storm water inputs to the river, will not only increase flow rates but also the level of suspended particles and, thereby, change the albedo on a daily and yearly basis. Together with radiation, evaporative heat transfer has also been shown to be important for the removal of heat from the river [12]. As the river temperature decreases due to evaporation the humidity of the surrounding area will increase and the higher temperature difference means the potential for sensible cooling of the air will also increase. Webb [12] found the evaporative heat flux varied significantly with weather conditions, increasing with high wind speeds and low humidity. Sensible heating of the river, which would directly relate to sensible cooling of the air, was found to have the greatest effects on the river temperatures in shaded areas, i.e. where there is limited impact from solar radiation. There are further interactions between the river, the river bed and the ground water. Although the work discussed above provides evidence for the heat transfer between water sources and the environment, and methods of modelling this behaviour, they do not provide direct evidence for the provision of cooling in a complex urban environment. Furthermore, these studies tend to focus on the impacts of river rehabilitation and consider the heat transfer processes in mainly rural locations; they do not focus on the effect of specific urban forms in the river corridor and the propagation of cooling from the river into the urban environment. Therefore, there is a need to assess the microclimate effects of urban rivers in the UK in order to evaluate their effectiveness in contributing to resilience to heat waves and how the urban form affects this. This research will focus specifically on small rivers. As well as representing a substantial number of urban watercourses smaller rivers are often representative of streams potentially being targeted for deculverting, or “forgotten” watercourses whose locality faces regeneration. Historically, many urban rivers in the UK were culverted (enclosed) as a consequence of urban expansion and the pressure to gain extra land in city centres. More recently, the daylighting of urban watercourses (i.e. the exposing of rivers) is being promoted by the UK’s Chartered Institute for Water and Environmental Management, who cite improvements to flood control and ecological benefits [14]; resilience to heat waves may be an added incentive. In order to aid decision making about the urban design in such areas it is necessary to develop a body of evidence for the potential such rivers have for cooling and the
15
impact different urban forms have on maintaining and propagating this effect. In this paper the temperature difference between an urban river, its surrounding and an urban reference site at distance from the river is examined during a spring/summer period to evaluate the potential cooling. The temperature difference is quantified based on the variation in ambient conditions and the urban form of the river corridor. Two periods with multiple consecutive days at high ambient air temperatures were also considered in more detail and used to assess the impact of the river temperature when ambient conditions were similar. 2. Methods The study is based along the River Don, Sheffield, UK, which flows with an average flow estimated to be 4.7 m3/s through the study site where the channel is approximately 22 m wide. The river passes through rural locations before entering the suburbs and finally the city of Sheffield. The UHI of the city has been measured as 2 C on a spring day [15]. Measurements of temperature and humidity were taken at 12 locations that were either directly adjacent to the river or running perpendicular to the river bank at a selection of sites close to the north of the city centre (see Fig. 1). Measurements were carried out with ibuttons (Maxim, USA) mounted in 12 plate Gill solar radiations screens (Skye Instruments). These have an accuracy of 0.5 C between 10 C and 65 C and a resolution of 0.0625 C. The loggers were calibrated before installation and recalibrated after a period of six months. These sites were all within 150 m of the city’s inner ring road. The loggers were all located at a height of 1.5 m above the ground, except at the two street sites (OStr and CStr) where they were mounted at a height of 3 m due to practical constraints (i.e. protection from damage). Measurements were downloaded every month and the internal clocks reset. An urban reference weather station (UR) (Skye Instruments, UK) was installed approximately 1.5 km from the site, at roof level, at a similar distance from the inner ring road as the study site. This monitored air temperature ( C), relative humidity (%), wind speed (m/s) and direction ( ) and solar radiation levels (W/m2). A second weather station located adjacent to the river (E2) at a height of 1.5 m monitored air temperature ( C), relative humidity (%), wind speed (m/s) and direction ( ) and water temperature ( C). Further comparative wind speed assessments were made at the two street sites using a vane anemometer (LCA501, TSI) for 30 min at 1 sec resolution. 2.1. Description of monitoring locations Distinct locations were chosen in order to assess the effect of different urban forms on the cooling effect of the river. Four different types of urban form were chosen: enclosed (E), open square (OSq), open street (OStr) and closed street (CStr). The layout of these sites are shown in Fig. 1 with the logger locations denoted by the letter code describing the type of urban form, and an individual number. Images of the sites are given in Fig. 2. There are two enclosed sites e at E2 the adjacent building is 6 storeys high, and at E1 the building is 7 storeys high. Both sites are on opposite sides of the river, enabling the microclimate effects to be assessed with differing building orientations. At the open square site the buildings are approximately 6 storeys high to the north and south, and 10 storeys high to the west. These surround an urban, paved square, approximately 25 50 m with small canopied trees in planters. The open street site is at a smaller scale than the other sites, with buildings only 2 storeys high to each side of a pedestrianised street 10 m wide opening out onto the river. The closed street site is directly adjacent and identical to the open street
16
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
Fig. 1. Diagram of the study site, showing the relevant location of the monitoring equipment.
except that the end of the street does not open out onto the water. It also has less shading than the open street section as shown in Fig. 2. Directly adjacent to the river at this site is a highly vegetated bank with mature trees. The conditions at these study sites were compared to those at the UR location. This reference site was chosen to be similar in all respects to the study sites except for its adjacency to water. The site is therefore located just inside the ring road, in order to represent the study site which spans the ring road. The area is built up, mainly hard paving and buildings 5 storeys high. To the north west are a series of towers 13 storeys high. The site is located approximately 1.5 km from the river. Due to the topography of the city it was necessary to locate the UR at an altitude 50 m higher than the study site. This is taken into account in the analysis. The weather station is located on the roof of a 5-storey building over a grassed green roof. This is similar to the planting under the majority of loggers located directly adjacent to the rivers and at the open square site.
2.2. Analysis This paper considers the effect of the river during warm weather, and therefore focuses on the period between the 24th April and 12th August 2010; monitor OS4 and E1 were not installed until the 18th June. In order to assess the temperature difference between the urban reference and each logger location, and therefore evaluate the potential cooling in proximity to the river, the mean difference between temperatures based on 20 min periods of data was calculated. The temperature differences between the urban reference and the particular logger Ti are calculated using:
DTi ¼ Ti TUR
(1)
where TUR is the temperature measured at the Urban Reference station, incorporating the lapse rate to take account of the altitude difference. The indices i is the reference for the individual loggers at the riverside. A negative value of DTi indicates cooling adjacent to the river in comparison to the urban reference.
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
17
Fig. 2. Images of each site typology. a) Open Square (OS), b) Open Street (OStr), c) Closed Street (CStr), d) Enclosed (E1), and e) Enclosed (E2).
Although the Environmental Lapse rate is generally given as 6.5 C/km up to 11 km above the Earth surface, this has been shown to vary widely, and has been found to be significantly lower closer to the Earth’s surface, reducing to 5 C/km within 2 km of sea level [16]. This value is also affected by local weather conditions: wind speed, solar radiation and humidity as well as the diurnal cycle. Measurements in the English uplands [17] have found significant variations in summer, with the lapse rate decreasing significantly overnight to only 0.2 C/km at 5am in summer and increasing to the dry adiabatic lapse rate of 9.81 C/km during the day. Due to the topography of the city the urban reference is 50 m higher than the study site, an altitude difference that can have a significant impact on measured temperature differences, but this will vary with the time of day and weather. Since artificially increasing the temperature at the reference location may positively impact the results a conservative value of 0.15 C was added to the measurements to generate TUR in order to account for the height difference. The values for DTi are calculated for the entire study, and then separately for all 20 min periods when TUR was greater than 15, 18, 20, 21, 22, 23 and 24 C. The period of time when cooling is required from the river is during hot weather, therefore, two individual periods during which the average air temperature was over 20 C for more than three consecutive days were analysed: May
20the24th and June 26theJuly 2nd, hereafter referred to as May and June respectively. Hourly average temperatures are calculated for both periods combined in order to understand the daily cooling range for periods with hot dry weather conditions. The two periods are then considered separately in order to evaluate the seasonal daytime and night time average differences in temperature at the river. The paired t-test [18] is used to assess the variance and statistical significance calculated using PASW statistics (IBM, USA). The mean difference (M) is plotted in the figures alongside the standard error (SE), which shows the variance in the sample means in order to convey the general trends. The statistical significance (p), tstatistic (t), degrees of freedom (df) and the effect size (r) are given in the main text or figure headings for a selection of key samples. The effect size in this instance is based on Pearson’s correlation coefficient [18]. Statistical significance was taken when p < 0.05 and an effect size >0.5 is considered to represent a large effect. Pearson’s correlation [19] is carried out on the influence of solar radiation, water temperature and wind speed on the results, presented as the correlation coefficient r, the statistical significance p, and the number of cases, n, where r values close to 1 represent a large positive correlation and r values close to 1 represent a large negative correlation.
18
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
3. Results 3.1. Role of ambient air temperature and urban form During the entire study period (24th Aprile12th August) TUR ranged between 1.7 C and 28.1 C. The wind speed varied from still to 7.2 m/s at the urban reference location, and reached a maximum of 2.5 m/s directly adjacent to the river. Figs. 3e5 shows the value of DTi for different ambient air temperature conditions at the different sites. Fig. 3a and b include all the loggers located directly adjacent to the river, the two charts show the results from the true right bank and the true left bank (facing downstream) separately, with the logger in the centre of the river (OS2) shown on both for reference. DTi varies with both urban form and ambient urban air temperature. As can be seen the loggers on the left generally do not show DTi outside the sensitivity of the logging equipment, except at the highly vegetated site. Moving away from the river’s edge the three site typologies e open square (Fig. 4), open street and closed street (Fig. 5) e are compared with DTi presented with increasing distance from the centre of river measured perpendicularly to the bank (5 m). Fig. 4 shows that for ambient temperatures greater than 20 C DTi is nearly 1 C over the river and statistically significant. The greatest measured cooling is over the river at OS2 (with a mean difference of 1.0 C (SE ¼ 0.43), t(1128) ¼ 23.1, p < 0.05, r ¼ 0.57), followed by OS3 with a mean difference of 0.7 C (SE ¼ 0.04), t(1127) ¼ 17.7, p < 0.05, r ¼ 0.47. Further from the river the effect size reduces rapidly and is no longer considered large 30 m
Fig. 3. Values of DTi for sites directly adjacent to the river. Shown for loggers on (a) the true right bank and (b) the true left bank. All filled markers are statistically significant p < 0.05, unfilled markers are not.
Fig. 4. Values of DTi moving away from the river at the open square site, with the distance from the river centre shown on the x-axis alongside the individual logger reference. All filled markers are statistically significant p < 0.05, unfilled markers are not.
away e at OS4 the mean difference is 0.4 C (SE ¼ 0.03), t(1128) ¼ 9.25, p < 0.05, r ¼ 0.27; at OS5 M ¼ 0.2 C, (SE ¼ 0.4),t(1128) ¼ 5.5, p < 0.05, r ¼ 0.16; and at OS6 M ¼ 0.6 C, (SE ¼ 0.04) t(1128) ¼ 15.4, p < 0.05, r ¼ 0.42. Similar results were found at the open streets site where the cooling decreased moving away from the river with the full
Fig. 5. Values of DTi moving away from the river at a) the closed street and b) the open streets. All filled markers are statistically significant p < 0.05, unfilled markers are not.
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
19
statistics shown in Table 1. The level of cooling is much greater on the open street than the closed at approximately 20 m from the river; however, these become similar 30 m away where the two streets join. 3.2. Consecutive periods >20 C The average TUR values during the two periods were 20.2 C (14.1e28.6 C) for May and 20.6 C (14.6e27.0 C) for June. The solar radiative flux on the site was also similar during the two periods with averages of 237 W/m2 in May and 241 W/m2 in June, as was the relative humidity at approximately 61%. However, the wind speed and water temperature differed for the two periods. In June the average wind speed was twice that in May, with a value of 1.9 m/s, compared to 0.9 m/s. The river was approximately two degrees warmer in June with an average temperature of 16.3 C compared to 14.6 C in May. Fig. 6aec shows the average DTi for each hour combining the two periods, with values shown separately for sites adjacent to the river bank, 30 m from the river and 40 m from the river. Very high levels of cooling are shown at the sites early in the morning, which reduces moving away from the river. The variation in cooling further from the river is particularly clear in the afternoon. Figs. 7 and 8 shows DTi above the river, and on the banks at the open square site, and full statistics are presented in Tables 2 and 3. The figures present the mean difference during the day (06:00e21:00) and night (21:00e06:00) separately and clearly show the cooling is only present during the daytime in both May and June. Cooling is shown in both months directly over the river; however, the magnitude is much lower in June, and reduces rapidly at the bank. The main differences in background conditions between these two periods are the solar angle, wind speed and water temperature. There is no comparison to rainfall events as the periods considered where completed dry. Correlations of DTOS2 to water temperature, solar radiation, wind speed and relative humidity are shown in Fig. 9aed respectively. Significant correlations for all four are found, with the greatest correlation being shown for solar radiation (r ¼ 0.714, p < 0.01, n ¼ 864), indicating an increasingly negative DTOS2, or cooling, above the river as solar radiation levels increase. The correlation to river water temperature gives a positive correlation showing the temperature at the riverside is warmer than the urban reference at high water temperatures (r ¼ 0.446, p < 0.01, n ¼ 864). Although Fig. 9(a) shows a large amount of scatter it demonstrates that periods when it is much warmer at the river than the urban reference are when the water is above 16 C. The correlation with relative humidity (r ¼ 0.369, p < 0.01, n ¼ 864), measured at the urban reference, shows greater cooling at the river with reduced humidity. The correlation to wind speed is also significant, although small (r ¼ 0.208, p < 0.01, n ¼ 864). Further analysis of the airflow through the open and closed streets was carried out based on measurements in each street. Airflow through the enclosed street was approximately half that through the open
Table 1 Mean DTi for the open and closed streets site when ambient temperatures >20 C. Also showing the t statistic (t), degrees and freedom (df) and effect size (r). Ref.
Mean D
SE
t
df
r
OStr1 OStr2 OStr3 CStr1 CStr2
1.2* 0.79* 0.34* 0.17* 0.24*
0.04 0.04 0.04 0.04 0.04
29.5 19.8 9.04 4.65 6.43
1128 1128 1128 1128 930
0.66 0.51 0.26 0.14 0.21
Tia
*P < 0.05. a Negative value represents cooling at riverside.
Fig. 6. Average hourly values of DTi for the May and June period combined at a) sites adjacent to the river, b) sites 30 m from the river and c) sites 40 m from the river.
street (average of 0.70 m/s in comparison 1.41 m/s) during a 30 min period when the reference measurement averaged 0.97 m/s. 4. Discussion 4.1. Use of lapse rate The measurements at the Urban Reference were increased by a value of 0.15 C in order to account for the altitude difference. However, as was discussed above, this value can vary with weather conditions. It is therefore prudent to ensure the patterns observed in the results are not due to the applied lapse rate. The lowest lapse rates as measured in the English Uplands were found to be during the night and early morning [17]. The results above show no
20
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22 Table 2 Mean DTi for daytime conditions during two warm periods. Also showing the t statistic (t), degrees and freedom (df) and effect size (r). Ref.
May Mean DTia
OS1 OS2 OS3 OS4 OS5 OS6
1.82* 1.53* 1.15* 1.03* 1.01*
June SE 0.27 0.25 0.25 0.28 0.20
t
df
6.64 6.17 4.52 3.64 5.16
24 24 24 24 24
r
Mean DTia
SE
t
df
r
0.80 0.78 0.68 0.60 0.73
0.26* 0.98* 0.69* 0.34 0.25 0.59*
0.14 0.20 0.18 0.18 0.20 0.18
1.79 4.82 3.77 1.86 1.27 3.33
34 34 34 34 34 34
0.29 0.64 0.54 0.30 0.21 0.50
*P < 0.05. a Negative value represents cooling at riverside.
Fig. 7. Values of DTi for the 20e24th May at the open square site moving away from the river, with the day and night shown separately. All filled markers are statistically significant (p > 0.95), unfilled markers are not.
significant cooling at this time, despite the addition of the lapse rate to the reference measurements. Therefore, if the 0.15 C chosen is too high it is not affecting the overall message given in the results. Although the use of the traditional higher value of 6.5 C/km would increase the presented cooling this would not seem realistic due to evidence that the lapse rate is significantly lower in the early morning. However, even if the lapse rate had been taken at the more common 6.5 C/km the majority of results would still fall within the accuracy of the equipment. The lapse rate has also been shown to increase with solar radiation. Since the results here found that cooling at the site increased with greater solar radiation, the use of a high value for the lapse rate would only have increased the values of cooling found at the site. Therefore, it seems reasonable that this choice of lapse rate takes some account of the altitude difference whilst being conservative in the presentation of results. 4.2. Role of ambient air temperature and urban form Cooling from the river is shown to vary with ambient air temperature, with greater cooling found at higher ambient air
temperatures at all sites. When TUR > 20 C, DTOS2 ¼ 1 C, showing a one degree difference in cooling at the river. The level of cooling on the bank was effected significantly by the local urban form with the greatest found on the highly vegetated bank (OStr1). The value of DTOStr1 is similar to DTOS2, which is located directly over the centre of the river projecting off a pedestrian footbridge. As such there is no vegetation directly adjacent to the logger, and the shading is much lower, only being affected by nearby buildings for short periods of the day in summer, and not being shaded from above, as is the case for OStr1. For the loggers directly adjacent to the river there is a clear difference between those sites on the true right, or left of the river. Those on the true right face approximately north east, and those on the true left face approximately south west (due to the curvature of the river this is not consistent). On the true right bank both sites show similar levels of cooling, with DTE1 w DTOS3 w 1.0 C with ambient temperatures over 24 C. However, on the true left bank this level of cooling is not found, expect for the heavily vegetated site. The cooling seems particularly limited at the sites which will receive substantial solar radiation, and in which the surrounding materials are dark brick or tarmac. This indicates the greater effect of the choice of materials, and the solar absorption on the site compared to the cooling available from the river. Further evidence that the treatment of the bank has significant effects on cooling is shown in Fig. 10, which presents the variation in cooling across the different sites for ambient temperature over 21 C. At all sites the level of cooling 40 m from the river is negligible. However, about 30 m from the bank there is considerable variation depending on the local urban form and provision of vegetation. Overall, the results indicate that high levels of vegetation next to the river increase the cooling on the bank, that opening up the streets to the river increases the propagation of cooling, and that the surface nature of the surrounding materials can have a more significant effect on the air temperatures than the presence of the river.
Table 3 Mean DTi for night time conditions during the two warm periods. Also showing the t statistic (t), degrees and freedom (df) and effect size (r). Ref.
May
June
Mean DTia SE
Fig. 8. Values of DTi for the 26th June e 2nd July at the open square site moving away from the river, with the day and night shown separately. All filled markers are statistically significant (p > 0.95), unfilled markers are not.
OS1 OS2 OS3 OS4 OS5 OS6
0.05 0.01 0.15 0.35* 0.08
0.14 0.15 0.15 0.15 0.13
t 0.38 0.09 1.00 2.35 0.65
df 13 13 13 13 13
r
Mean DTia SE
t
df
r
0.10 0.02 0.27 0.55 0.18
0.66* 0.77 0.81 0.91 1.10 1.03
4.98 5.22 5.24 5.70 7.16 7.22
20 20 20 20 20 20
0.74 1.94 1.92 1.61 1.28 1.27
*P < 0.05. a Negative value represents cooling at riverside.
0.14 0.15 0.15 0.16 0.15 0.14
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
21
Fig. 9. Scatter plots of DTOS2 and a) river water temperature, b) solar radiation, c) wind speed and d) relative humidity.
4.3. Consecutive periods >20 C Consideration of sustained periods of hot weather show cooling at the river only occurs in the daytime (Figs. 6 and 7), which agrees with the results of Murakawa et al. [11]. Murakawa studied periods with much greater ambient air temperatures, showing cooling directly above the river in the day. However, after sundown the temperature differences across the site reduced. This has implications for building design, particularly those building employing
Fig. 10. Values of DTi moving away from the river at the three different sites for ambient temperatures >21 C. All filled markers are statistically significant (p > 0.95).
night time ventilation cooling. If such a strategy was incorporated in a building adjacent to a river there would be little benefit from locating air inlets close to the river to improve night cooling, although there would be benefits for reducing daytime peaks in temperature. The cooling in the day is particularly high during the morning periods. Even at this time there is substantial variation, dependent on the local urban forms, of approximately 2 C. However, during the morning periods the complex nature of the urban site means aspects such as shading may have a significant effect when the sun is low. Therefore these high levels of cooling should be viewed with caution. However, at midday when implications of shading from surrounding buildings should be reduced there is still over 1 C cooling at all sites near the river expect the closed street site. Although some cooling is shown in the daytime during both periods there are substantial differences between the cooling shown at the river during the daytime in May and in June, which may be due to a variety of factors. Consideration of the 24 h data showed solar radiative flux had a strong negative correlation with DTOS2, indicating greater cooling with higher solar radiative flux. However, since the average for the two periods was similar it is likely this effect is related to the observed diurnal variations in cooling rather than seasonal differences. The next greatest correlation with DTOS2 was water temperature. Since the average water temperature in June is approximately 2 C warmer than in May this may impact on the cooling potential as the warmer river temperatures will reduce the availability of sensible cooling. For small urban rivers dry periods may result in very shallow levels of flow in the river. This reduction in volume means the water will increase in temperature more rapidly, and at higher water temperatures there
22
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
is a reduced capacity for sensible cooling. The River Don is a relatively small river, with a mean flow rate approximately thirteen times smaller than the River Thames as it passes through Kingston [20]. For both these periods the Don’s flow was low: 1.4 m3/s in June and 2.12 m3/s in May. Therefore these results are likely to be representative for smaller urban rivers, but not large water bodies with a more constant yearly temperature. However, further work would be needed to assess this. It is also worth noting that industrial discharges can result in higher water temperatures which will also reduce the capacity for cooling. Even with cooling only being provided in the spring, when the water temperature is cooler, this can have a substantial benefit on comfort. The lack of phased adaptation to the warmer temperatures means spring time heat waves can create greater relative discomfort than in the summer, and so the presence of a water body providing cooling at this time can be highly beneficial. Wind speed may also have implications for different levels of cooling measured at the site since the wind speed was double in June what it was in May. Although higher wind speeds should promote evaporation and, therefore, cooling of the river and potentially the air, this study has shown temperatures at the site more similar to the urban fabric during periods of high wind speeds. Smaller or positive values of DTi do not necessarily represent less cooling from the river, as the higher wind speeds will remove the air from the site more rapidly, and result in greater mixing of the air. Therefore, the correlation between wind speed and positive values of DTi may be due to warmer air being transported to the site. 4.4. Practical implications for the urban form in river corridors The impact of the urban form on the results show the importance of the design of the river corridor, over the simple presence of the water way. The results show that general bioclimatic design principles hold true in the design of river corridors. Greatest cooling is found where there is high amount of tree cover and shading. Any benefits of the river are lost quickly where the site is physically cut off from the water by walls or buildings. Cooling also quickly diminishes when the bankside consists of dark materials such as brick or tarmac with a south west orientation. 5. Conclusions This research presents the results from the study of the microclimate around a small urban river in the UK, representative of many other watercourses in built up areas. Due to the length of the study substantial variations in weather conditions can be considered, alongside the relative impacts of different types of urban typology. The results show significant cooling over the river with an average cooling of nearly 1 C during ambient conditions greater than 20 C. For consecutive hot days in May there was substantial cooling shown over the river of nearly 2 C and 1.5 C cooling on the river bank. This was only shown during the day and was not evident for night time temperature contours across the whole site. In late June, with a similar ambient air temperature, the cooling had diminished. Hourly averages of the two hot periods showed that this cooling tended to be greatest in the mornings with cooling at the river bank varying by 2 C depending on the urban form. Therefore in order to maximise the benefits of this cooling close consideration of the urban design is required. Highly vegetated banks showed much lower temperatures than those banks consisting of only hard
engineering materials. The nature of this surrounding material can have a greater impact on air temperatures than the presence of the river alone. Opening up the streets to the river or the provision of a square gave greater cooling 30 m from the bank than streets shut off from the river. However it was clear that other microclimate factors, such as material albedo and shading, may have been an influence. Therefore, further work into these aspects is required to provide detailed design guidance to gain the most effective cooling from an urban river. Acknowledgements This paper is based on work undertaken as part of the URSULA project funded by the Engineering and Physical Sciences Research Council (grant number EP/F007388/1). The authors are grateful for EPSRC’s support. The views presented in the paper are those of the authors, and cannot be taken as indicative in any way of the position of URSULA colleagues, partners or of EPSRC. All errors are those of the authors alone. References [1] Smith C, Levermore G. Designing urban spaces and buildings to improve sustainability and quality of life in a warmer world. Energ Pol 2008;36: 4558e62. [2] Upmani H, Eliasson I, Lindqvist S. The influence of green areas on nocturnal temperatures in a high latitude city (Göteborg, Sweden). Int J Climatol 1998; 18:681e700. [3] Jansson C. Urban microclimate and surface hydrometeorological processes, thesis. [Stockholm]: Royal Institute of Technology (KTH). 2006. p. 24. [4] Jauregui E. Influence of a large urban park on temperature and convective precipitation in a tropical city. Energy Build 1990;15e16:457e63. [5] Shashua-Bar L, Hoffman ME. Vegetation as a climatic component in the design of an urban street: an empirical model for predicting the cooling effect of urban green areas with trees. Energy Build 2000;31:221e35. [6] Yang F, Lau SSY, Qian F. Summer time heat island intensities in three high rise housing quarters in inner city Shanghai, China. Build Environ 2010;45: 115e34. [7] Taha H. Urban climates and heat islands: albedo, evapotranspiration and anthropogenic heat. Energy Build 1997;25:99e103. [8] Grimmond CSB, Oke TR. An evapotranspiration-interception model for urban areas. Water Resour Res 1991;277:1739e55. [9] Kim YH. Does the restoration of an inner-city stream in Seoul affect local thermal environment. Theor Appl Climatol 2008;92:239e48. [10] Huang l, Zhao D, Wang J, Zhu J, Li J. Scale impacts of land cover and vegetation corridors on urban thermal behavior in Nanjing, China. Theor Appl Climatol 2007;94:241e57. [11] Murakawa S, Sekine T, Narita K. Study of the effects of a river on the thermal environment in an urban area. Energy Build 1990;15:993e1001. [12] Webb BW, Zhang Y. Spatial and seasonal variability in the components of the river heat budget. Hydrol Process 1997;11:79e101. [13] Hannah DM, Malcolm IA, Soulsby C, Youngson AF. Heat exchanges and temperatures within a salmon spawning stream in the Cairngorms, Scotland: seasonal and sub-seasonal dynamics. River Res Appl 2004;20:635e52. [14] CIWEM. Policy position statement on deculverting of watercourses [Internet]. London: Chartered Institute of Water and Environmental Management [updated Dec 2007: cited 2012 Feb 15]. Available from: http://www.ciwem. org/policy-and-international/policy-position-statements; 2007. [15] Lee SE, Sharples S. An analysis of the Urban Heat Island of Sheffield e the impact of a changing climate. In: Kenny P, Brophy V, Lewis JO, editors. Proceedings of the 25th conference on passive and low energy architecture. Dublin: University College Dublin; 2008. p. 396. [16] Barry RG. Atmosphere, weather and climate. 9th ed. London: Routledge; 2009. [17] Pepin N, Bentham D, Taylor K. Modelling lapse rates in maritime uplands of northern England: implications for climate change. Artic Antartic Alpine Res 1999;31:151e64. [18] Rosnow RL, Rosenthal R, Rubin DB. Contrasts and correlations in effect size estimations. Psychol Sci 2000;11:446e53. [19] Hinton P. Statistics explained. 2nd ed. London: Routledge; 2004. [20] CEH. National river flow archive [Internet]. Wallingford (UK): NERC e Centre for Ecology and Hydrology [Updated 2012; cited 2012 Feb 15] Available from: http://www.ceh.ac.uk/data/nrfa/data/search.html; 2010.
Contents lists available at SciVerse ScienceDirect
Building and Environment journal homepage: www.elsevier.com/locate/buildenv
The interaction of rivers and urban form in mitigating the Urban Heat Island effect: A UK case study E.A. Hathway a, *, S. Sharples b a b
Department of Civil and Structural Engineering, University of Sheffield, Sir Frederick Mappin Building, Mappin Street, Sheffield, South Yorkshire S1 3JD, UK School of Architecture, University of Liverpool, Liverpool, UK
a r t i c l e i n f o
a b s t r a c t
Article history: Received 16 March 2012 Received in revised form 19 June 2012 Accepted 20 June 2012
The Urban Heat Island (UHI) effect already produces elevated temperatures in city centres therefore urban design has a key role to pay in reducing the UHI to create safe and pleasant places in which to live and work. Increased surface porosity and bodies of surface water have a role to play in increasing potential cooling through evaporation. Urban rivers may, therefore, have a place in reducing the UHI. This paper investigates the effectiveness that small urban rivers may have in reducing the UHI effect and also examines the role that the urban form on the banks of a river can play in propagating or reducing this potential cooling. The results from a field survey during spring and summer are presented for a river in Sheffield, UK. The level of cooling is related to the ambient air temperature, increasing at higher temperatures. However, there are also seasonal dependencies and relationships linked to the river water temperature, incident solar radiation, wind speed and relative humidity. A mean level of daytime cooling of over 1.5 C was found above the river in spring, but this was reduced in summer when the river water temperature was warmer. The urban form on the river bank influenced the levels of cooling felt away from the river bank. Ó 2012 Elsevier Ltd. All rights reserved.
Keywords: Urban Heat Island Rivers Microclimate Blue infrastructure
1. Introduction It is becoming increasingly important to provide resilience to the urban infrastructure in order to withstand extreme events brought about by a changing climate. Flooding, high wind speeds and increased air temperatures all present different risks to cities. Increases in mean air temperatures and more frequent heat waves means the Urban Heat Island (UHI) is becoming an increasingly significant problem in the UK. An overview of the impacts of the UHI in the UK and methods for mitigation and climate change resilience has been reviewed by Smith and Levermore [1]. Globally, there is a significant body of research attempting to understand and quantify local microclimates, often with the aim of providing design guidance to generate better quality urban spaces. Of particular interest is the provision of vegetation and green spaces; both large and small parks have been found to provide cooling, with the effects propagating to a distance approximately half the park width away, dependent upon the local street layout [2e5]. The provision of large canopy street trees can offset the heat input from vehicles [6] and the choice of built environment surfaces with high solar
* Corresponding author. Tel.: þ44 (0) 114 2225702; fax: þ44 (0) 114 2225700. E-mail address: a.hathway@sheffield.ac.uk (E.A. Hathway). 0360-1323/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.buildenv.2012.06.013
reflectances (albedo) can result in lower levels of solar absorption and so cooler surfaces and adjacent air temperatures [7]. Alongside the provision of shading and cooling via evapotranspiration from vegetation, green spaces will also usually improve the surface porosity, thereby increasing the available capacity for water storage and so water availability for evaporative cooling. The reintroduction of water through the deliberate incorporation of porous surfaces, e.g. porous paving, or the presence of water bodies, such as ponds or rivers, has the potential to reduce the UHI by returning the surface moisture availability to values similar to rural areas. The process of evaporation has been studied, and resulting cooling for various cities demonstrated with models validated for a range of locations [8]. However, there is limited published research on the microclimate effects in the immediate locality of river corridors, with current published data limited to tropical climates. The process of daylighting a large stretch of watercourse in Seoul, Korea, provided the opportunity of a before and after study demonstrating cooling of the urban microclimate [9]. Another study in Nanjing, China [10] of the urban climate highlighted the cooling effects in the locality of a lake, river and sea. The cooling effect of the Ota River, Hiroshima, Japan was found to reach up to 5 C directly above the river and propagated nearly 100 m from the river banks [11]. These studies demonstrate the potential cooling from rivers in hot climates. Further analysis of the
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
influence of local urban form on the propagation of this cooling would be beneficial for providing guidance on river corridor regeneration for an improved microclimate. Although there are limited published studies on the cooling effects rivers provide in Europe, the heat flux between the atmosphere and rivers has been studied extensively in relation to the thermal properties of the river water. This is mainly due to the resulting ecological implications such as fish spawning [12,13]. The heat budget of a river is related to the sensible, radiative and evaporative heat fluxes between it and the surroundings. The process resulting in the greatest effect is dependent upon the local environment. The direction of heat flux varies during the diurnal cycle, for instance absorbing solar radiation during the day, and radiating long wave radiation at night. However, it seems the radiative heat attenuated and absorbed by the river during the day is greater than that released to the environment during the evening and night, at least for rural areas [12]. In comparison to static absorption and release of radiation for most urban materials, the heat absorbed by the water and suspended particles will be carried downstream at a rate dependent on the river dynamics; heat absorbed will then be released at a different location downstream. The albedo of river water also varies with the angle of incidence of the solar radiation and the quantity of suspended particles. Therefore, intense rain events, resulting in large storm water inputs to the river, will not only increase flow rates but also the level of suspended particles and, thereby, change the albedo on a daily and yearly basis. Together with radiation, evaporative heat transfer has also been shown to be important for the removal of heat from the river [12]. As the river temperature decreases due to evaporation the humidity of the surrounding area will increase and the higher temperature difference means the potential for sensible cooling of the air will also increase. Webb [12] found the evaporative heat flux varied significantly with weather conditions, increasing with high wind speeds and low humidity. Sensible heating of the river, which would directly relate to sensible cooling of the air, was found to have the greatest effects on the river temperatures in shaded areas, i.e. where there is limited impact from solar radiation. There are further interactions between the river, the river bed and the ground water. Although the work discussed above provides evidence for the heat transfer between water sources and the environment, and methods of modelling this behaviour, they do not provide direct evidence for the provision of cooling in a complex urban environment. Furthermore, these studies tend to focus on the impacts of river rehabilitation and consider the heat transfer processes in mainly rural locations; they do not focus on the effect of specific urban forms in the river corridor and the propagation of cooling from the river into the urban environment. Therefore, there is a need to assess the microclimate effects of urban rivers in the UK in order to evaluate their effectiveness in contributing to resilience to heat waves and how the urban form affects this. This research will focus specifically on small rivers. As well as representing a substantial number of urban watercourses smaller rivers are often representative of streams potentially being targeted for deculverting, or “forgotten” watercourses whose locality faces regeneration. Historically, many urban rivers in the UK were culverted (enclosed) as a consequence of urban expansion and the pressure to gain extra land in city centres. More recently, the daylighting of urban watercourses (i.e. the exposing of rivers) is being promoted by the UK’s Chartered Institute for Water and Environmental Management, who cite improvements to flood control and ecological benefits [14]; resilience to heat waves may be an added incentive. In order to aid decision making about the urban design in such areas it is necessary to develop a body of evidence for the potential such rivers have for cooling and the
15
impact different urban forms have on maintaining and propagating this effect. In this paper the temperature difference between an urban river, its surrounding and an urban reference site at distance from the river is examined during a spring/summer period to evaluate the potential cooling. The temperature difference is quantified based on the variation in ambient conditions and the urban form of the river corridor. Two periods with multiple consecutive days at high ambient air temperatures were also considered in more detail and used to assess the impact of the river temperature when ambient conditions were similar. 2. Methods The study is based along the River Don, Sheffield, UK, which flows with an average flow estimated to be 4.7 m3/s through the study site where the channel is approximately 22 m wide. The river passes through rural locations before entering the suburbs and finally the city of Sheffield. The UHI of the city has been measured as 2 C on a spring day [15]. Measurements of temperature and humidity were taken at 12 locations that were either directly adjacent to the river or running perpendicular to the river bank at a selection of sites close to the north of the city centre (see Fig. 1). Measurements were carried out with ibuttons (Maxim, USA) mounted in 12 plate Gill solar radiations screens (Skye Instruments). These have an accuracy of 0.5 C between 10 C and 65 C and a resolution of 0.0625 C. The loggers were calibrated before installation and recalibrated after a period of six months. These sites were all within 150 m of the city’s inner ring road. The loggers were all located at a height of 1.5 m above the ground, except at the two street sites (OStr and CStr) where they were mounted at a height of 3 m due to practical constraints (i.e. protection from damage). Measurements were downloaded every month and the internal clocks reset. An urban reference weather station (UR) (Skye Instruments, UK) was installed approximately 1.5 km from the site, at roof level, at a similar distance from the inner ring road as the study site. This monitored air temperature ( C), relative humidity (%), wind speed (m/s) and direction ( ) and solar radiation levels (W/m2). A second weather station located adjacent to the river (E2) at a height of 1.5 m monitored air temperature ( C), relative humidity (%), wind speed (m/s) and direction ( ) and water temperature ( C). Further comparative wind speed assessments were made at the two street sites using a vane anemometer (LCA501, TSI) for 30 min at 1 sec resolution. 2.1. Description of monitoring locations Distinct locations were chosen in order to assess the effect of different urban forms on the cooling effect of the river. Four different types of urban form were chosen: enclosed (E), open square (OSq), open street (OStr) and closed street (CStr). The layout of these sites are shown in Fig. 1 with the logger locations denoted by the letter code describing the type of urban form, and an individual number. Images of the sites are given in Fig. 2. There are two enclosed sites e at E2 the adjacent building is 6 storeys high, and at E1 the building is 7 storeys high. Both sites are on opposite sides of the river, enabling the microclimate effects to be assessed with differing building orientations. At the open square site the buildings are approximately 6 storeys high to the north and south, and 10 storeys high to the west. These surround an urban, paved square, approximately 25 50 m with small canopied trees in planters. The open street site is at a smaller scale than the other sites, with buildings only 2 storeys high to each side of a pedestrianised street 10 m wide opening out onto the river. The closed street site is directly adjacent and identical to the open street
16
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
Fig. 1. Diagram of the study site, showing the relevant location of the monitoring equipment.
except that the end of the street does not open out onto the water. It also has less shading than the open street section as shown in Fig. 2. Directly adjacent to the river at this site is a highly vegetated bank with mature trees. The conditions at these study sites were compared to those at the UR location. This reference site was chosen to be similar in all respects to the study sites except for its adjacency to water. The site is therefore located just inside the ring road, in order to represent the study site which spans the ring road. The area is built up, mainly hard paving and buildings 5 storeys high. To the north west are a series of towers 13 storeys high. The site is located approximately 1.5 km from the river. Due to the topography of the city it was necessary to locate the UR at an altitude 50 m higher than the study site. This is taken into account in the analysis. The weather station is located on the roof of a 5-storey building over a grassed green roof. This is similar to the planting under the majority of loggers located directly adjacent to the rivers and at the open square site.
2.2. Analysis This paper considers the effect of the river during warm weather, and therefore focuses on the period between the 24th April and 12th August 2010; monitor OS4 and E1 were not installed until the 18th June. In order to assess the temperature difference between the urban reference and each logger location, and therefore evaluate the potential cooling in proximity to the river, the mean difference between temperatures based on 20 min periods of data was calculated. The temperature differences between the urban reference and the particular logger Ti are calculated using:
DTi ¼ Ti TUR
(1)
where TUR is the temperature measured at the Urban Reference station, incorporating the lapse rate to take account of the altitude difference. The indices i is the reference for the individual loggers at the riverside. A negative value of DTi indicates cooling adjacent to the river in comparison to the urban reference.
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
17
Fig. 2. Images of each site typology. a) Open Square (OS), b) Open Street (OStr), c) Closed Street (CStr), d) Enclosed (E1), and e) Enclosed (E2).
Although the Environmental Lapse rate is generally given as 6.5 C/km up to 11 km above the Earth surface, this has been shown to vary widely, and has been found to be significantly lower closer to the Earth’s surface, reducing to 5 C/km within 2 km of sea level [16]. This value is also affected by local weather conditions: wind speed, solar radiation and humidity as well as the diurnal cycle. Measurements in the English uplands [17] have found significant variations in summer, with the lapse rate decreasing significantly overnight to only 0.2 C/km at 5am in summer and increasing to the dry adiabatic lapse rate of 9.81 C/km during the day. Due to the topography of the city the urban reference is 50 m higher than the study site, an altitude difference that can have a significant impact on measured temperature differences, but this will vary with the time of day and weather. Since artificially increasing the temperature at the reference location may positively impact the results a conservative value of 0.15 C was added to the measurements to generate TUR in order to account for the height difference. The values for DTi are calculated for the entire study, and then separately for all 20 min periods when TUR was greater than 15, 18, 20, 21, 22, 23 and 24 C. The period of time when cooling is required from the river is during hot weather, therefore, two individual periods during which the average air temperature was over 20 C for more than three consecutive days were analysed: May
20the24th and June 26theJuly 2nd, hereafter referred to as May and June respectively. Hourly average temperatures are calculated for both periods combined in order to understand the daily cooling range for periods with hot dry weather conditions. The two periods are then considered separately in order to evaluate the seasonal daytime and night time average differences in temperature at the river. The paired t-test [18] is used to assess the variance and statistical significance calculated using PASW statistics (IBM, USA). The mean difference (M) is plotted in the figures alongside the standard error (SE), which shows the variance in the sample means in order to convey the general trends. The statistical significance (p), tstatistic (t), degrees of freedom (df) and the effect size (r) are given in the main text or figure headings for a selection of key samples. The effect size in this instance is based on Pearson’s correlation coefficient [18]. Statistical significance was taken when p < 0.05 and an effect size >0.5 is considered to represent a large effect. Pearson’s correlation [19] is carried out on the influence of solar radiation, water temperature and wind speed on the results, presented as the correlation coefficient r, the statistical significance p, and the number of cases, n, where r values close to 1 represent a large positive correlation and r values close to 1 represent a large negative correlation.
18
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
3. Results 3.1. Role of ambient air temperature and urban form During the entire study period (24th Aprile12th August) TUR ranged between 1.7 C and 28.1 C. The wind speed varied from still to 7.2 m/s at the urban reference location, and reached a maximum of 2.5 m/s directly adjacent to the river. Figs. 3e5 shows the value of DTi for different ambient air temperature conditions at the different sites. Fig. 3a and b include all the loggers located directly adjacent to the river, the two charts show the results from the true right bank and the true left bank (facing downstream) separately, with the logger in the centre of the river (OS2) shown on both for reference. DTi varies with both urban form and ambient urban air temperature. As can be seen the loggers on the left generally do not show DTi outside the sensitivity of the logging equipment, except at the highly vegetated site. Moving away from the river’s edge the three site typologies e open square (Fig. 4), open street and closed street (Fig. 5) e are compared with DTi presented with increasing distance from the centre of river measured perpendicularly to the bank (5 m). Fig. 4 shows that for ambient temperatures greater than 20 C DTi is nearly 1 C over the river and statistically significant. The greatest measured cooling is over the river at OS2 (with a mean difference of 1.0 C (SE ¼ 0.43), t(1128) ¼ 23.1, p < 0.05, r ¼ 0.57), followed by OS3 with a mean difference of 0.7 C (SE ¼ 0.04), t(1127) ¼ 17.7, p < 0.05, r ¼ 0.47. Further from the river the effect size reduces rapidly and is no longer considered large 30 m
Fig. 3. Values of DTi for sites directly adjacent to the river. Shown for loggers on (a) the true right bank and (b) the true left bank. All filled markers are statistically significant p < 0.05, unfilled markers are not.
Fig. 4. Values of DTi moving away from the river at the open square site, with the distance from the river centre shown on the x-axis alongside the individual logger reference. All filled markers are statistically significant p < 0.05, unfilled markers are not.
away e at OS4 the mean difference is 0.4 C (SE ¼ 0.03), t(1128) ¼ 9.25, p < 0.05, r ¼ 0.27; at OS5 M ¼ 0.2 C, (SE ¼ 0.4),t(1128) ¼ 5.5, p < 0.05, r ¼ 0.16; and at OS6 M ¼ 0.6 C, (SE ¼ 0.04) t(1128) ¼ 15.4, p < 0.05, r ¼ 0.42. Similar results were found at the open streets site where the cooling decreased moving away from the river with the full
Fig. 5. Values of DTi moving away from the river at a) the closed street and b) the open streets. All filled markers are statistically significant p < 0.05, unfilled markers are not.
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
19
statistics shown in Table 1. The level of cooling is much greater on the open street than the closed at approximately 20 m from the river; however, these become similar 30 m away where the two streets join. 3.2. Consecutive periods >20 C The average TUR values during the two periods were 20.2 C (14.1e28.6 C) for May and 20.6 C (14.6e27.0 C) for June. The solar radiative flux on the site was also similar during the two periods with averages of 237 W/m2 in May and 241 W/m2 in June, as was the relative humidity at approximately 61%. However, the wind speed and water temperature differed for the two periods. In June the average wind speed was twice that in May, with a value of 1.9 m/s, compared to 0.9 m/s. The river was approximately two degrees warmer in June with an average temperature of 16.3 C compared to 14.6 C in May. Fig. 6aec shows the average DTi for each hour combining the two periods, with values shown separately for sites adjacent to the river bank, 30 m from the river and 40 m from the river. Very high levels of cooling are shown at the sites early in the morning, which reduces moving away from the river. The variation in cooling further from the river is particularly clear in the afternoon. Figs. 7 and 8 shows DTi above the river, and on the banks at the open square site, and full statistics are presented in Tables 2 and 3. The figures present the mean difference during the day (06:00e21:00) and night (21:00e06:00) separately and clearly show the cooling is only present during the daytime in both May and June. Cooling is shown in both months directly over the river; however, the magnitude is much lower in June, and reduces rapidly at the bank. The main differences in background conditions between these two periods are the solar angle, wind speed and water temperature. There is no comparison to rainfall events as the periods considered where completed dry. Correlations of DTOS2 to water temperature, solar radiation, wind speed and relative humidity are shown in Fig. 9aed respectively. Significant correlations for all four are found, with the greatest correlation being shown for solar radiation (r ¼ 0.714, p < 0.01, n ¼ 864), indicating an increasingly negative DTOS2, or cooling, above the river as solar radiation levels increase. The correlation to river water temperature gives a positive correlation showing the temperature at the riverside is warmer than the urban reference at high water temperatures (r ¼ 0.446, p < 0.01, n ¼ 864). Although Fig. 9(a) shows a large amount of scatter it demonstrates that periods when it is much warmer at the river than the urban reference are when the water is above 16 C. The correlation with relative humidity (r ¼ 0.369, p < 0.01, n ¼ 864), measured at the urban reference, shows greater cooling at the river with reduced humidity. The correlation to wind speed is also significant, although small (r ¼ 0.208, p < 0.01, n ¼ 864). Further analysis of the airflow through the open and closed streets was carried out based on measurements in each street. Airflow through the enclosed street was approximately half that through the open
Table 1 Mean DTi for the open and closed streets site when ambient temperatures >20 C. Also showing the t statistic (t), degrees and freedom (df) and effect size (r). Ref.
Mean D
SE
t
df
r
OStr1 OStr2 OStr3 CStr1 CStr2
1.2* 0.79* 0.34* 0.17* 0.24*
0.04 0.04 0.04 0.04 0.04
29.5 19.8 9.04 4.65 6.43
1128 1128 1128 1128 930
0.66 0.51 0.26 0.14 0.21
Tia
*P < 0.05. a Negative value represents cooling at riverside.
Fig. 6. Average hourly values of DTi for the May and June period combined at a) sites adjacent to the river, b) sites 30 m from the river and c) sites 40 m from the river.
street (average of 0.70 m/s in comparison 1.41 m/s) during a 30 min period when the reference measurement averaged 0.97 m/s. 4. Discussion 4.1. Use of lapse rate The measurements at the Urban Reference were increased by a value of 0.15 C in order to account for the altitude difference. However, as was discussed above, this value can vary with weather conditions. It is therefore prudent to ensure the patterns observed in the results are not due to the applied lapse rate. The lowest lapse rates as measured in the English Uplands were found to be during the night and early morning [17]. The results above show no
20
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22 Table 2 Mean DTi for daytime conditions during two warm periods. Also showing the t statistic (t), degrees and freedom (df) and effect size (r). Ref.
May Mean DTia
OS1 OS2 OS3 OS4 OS5 OS6
1.82* 1.53* 1.15* 1.03* 1.01*
June SE 0.27 0.25 0.25 0.28 0.20
t
df
6.64 6.17 4.52 3.64 5.16
24 24 24 24 24
r
Mean DTia
SE
t
df
r
0.80 0.78 0.68 0.60 0.73
0.26* 0.98* 0.69* 0.34 0.25 0.59*
0.14 0.20 0.18 0.18 0.20 0.18
1.79 4.82 3.77 1.86 1.27 3.33
34 34 34 34 34 34
0.29 0.64 0.54 0.30 0.21 0.50
*P < 0.05. a Negative value represents cooling at riverside.
Fig. 7. Values of DTi for the 20e24th May at the open square site moving away from the river, with the day and night shown separately. All filled markers are statistically significant (p > 0.95), unfilled markers are not.
significant cooling at this time, despite the addition of the lapse rate to the reference measurements. Therefore, if the 0.15 C chosen is too high it is not affecting the overall message given in the results. Although the use of the traditional higher value of 6.5 C/km would increase the presented cooling this would not seem realistic due to evidence that the lapse rate is significantly lower in the early morning. However, even if the lapse rate had been taken at the more common 6.5 C/km the majority of results would still fall within the accuracy of the equipment. The lapse rate has also been shown to increase with solar radiation. Since the results here found that cooling at the site increased with greater solar radiation, the use of a high value for the lapse rate would only have increased the values of cooling found at the site. Therefore, it seems reasonable that this choice of lapse rate takes some account of the altitude difference whilst being conservative in the presentation of results. 4.2. Role of ambient air temperature and urban form Cooling from the river is shown to vary with ambient air temperature, with greater cooling found at higher ambient air
temperatures at all sites. When TUR > 20 C, DTOS2 ¼ 1 C, showing a one degree difference in cooling at the river. The level of cooling on the bank was effected significantly by the local urban form with the greatest found on the highly vegetated bank (OStr1). The value of DTOStr1 is similar to DTOS2, which is located directly over the centre of the river projecting off a pedestrian footbridge. As such there is no vegetation directly adjacent to the logger, and the shading is much lower, only being affected by nearby buildings for short periods of the day in summer, and not being shaded from above, as is the case for OStr1. For the loggers directly adjacent to the river there is a clear difference between those sites on the true right, or left of the river. Those on the true right face approximately north east, and those on the true left face approximately south west (due to the curvature of the river this is not consistent). On the true right bank both sites show similar levels of cooling, with DTE1 w DTOS3 w 1.0 C with ambient temperatures over 24 C. However, on the true left bank this level of cooling is not found, expect for the heavily vegetated site. The cooling seems particularly limited at the sites which will receive substantial solar radiation, and in which the surrounding materials are dark brick or tarmac. This indicates the greater effect of the choice of materials, and the solar absorption on the site compared to the cooling available from the river. Further evidence that the treatment of the bank has significant effects on cooling is shown in Fig. 10, which presents the variation in cooling across the different sites for ambient temperature over 21 C. At all sites the level of cooling 40 m from the river is negligible. However, about 30 m from the bank there is considerable variation depending on the local urban form and provision of vegetation. Overall, the results indicate that high levels of vegetation next to the river increase the cooling on the bank, that opening up the streets to the river increases the propagation of cooling, and that the surface nature of the surrounding materials can have a more significant effect on the air temperatures than the presence of the river.
Table 3 Mean DTi for night time conditions during the two warm periods. Also showing the t statistic (t), degrees and freedom (df) and effect size (r). Ref.
May
June
Mean DTia SE
Fig. 8. Values of DTi for the 26th June e 2nd July at the open square site moving away from the river, with the day and night shown separately. All filled markers are statistically significant (p > 0.95), unfilled markers are not.
OS1 OS2 OS3 OS4 OS5 OS6
0.05 0.01 0.15 0.35* 0.08
0.14 0.15 0.15 0.15 0.13
t 0.38 0.09 1.00 2.35 0.65
df 13 13 13 13 13
r
Mean DTia SE
t
df
r
0.10 0.02 0.27 0.55 0.18
0.66* 0.77 0.81 0.91 1.10 1.03
4.98 5.22 5.24 5.70 7.16 7.22
20 20 20 20 20 20
0.74 1.94 1.92 1.61 1.28 1.27
*P < 0.05. a Negative value represents cooling at riverside.
0.14 0.15 0.15 0.16 0.15 0.14
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
21
Fig. 9. Scatter plots of DTOS2 and a) river water temperature, b) solar radiation, c) wind speed and d) relative humidity.
4.3. Consecutive periods >20 C Consideration of sustained periods of hot weather show cooling at the river only occurs in the daytime (Figs. 6 and 7), which agrees with the results of Murakawa et al. [11]. Murakawa studied periods with much greater ambient air temperatures, showing cooling directly above the river in the day. However, after sundown the temperature differences across the site reduced. This has implications for building design, particularly those building employing
Fig. 10. Values of DTi moving away from the river at the three different sites for ambient temperatures >21 C. All filled markers are statistically significant (p > 0.95).
night time ventilation cooling. If such a strategy was incorporated in a building adjacent to a river there would be little benefit from locating air inlets close to the river to improve night cooling, although there would be benefits for reducing daytime peaks in temperature. The cooling in the day is particularly high during the morning periods. Even at this time there is substantial variation, dependent on the local urban forms, of approximately 2 C. However, during the morning periods the complex nature of the urban site means aspects such as shading may have a significant effect when the sun is low. Therefore these high levels of cooling should be viewed with caution. However, at midday when implications of shading from surrounding buildings should be reduced there is still over 1 C cooling at all sites near the river expect the closed street site. Although some cooling is shown in the daytime during both periods there are substantial differences between the cooling shown at the river during the daytime in May and in June, which may be due to a variety of factors. Consideration of the 24 h data showed solar radiative flux had a strong negative correlation with DTOS2, indicating greater cooling with higher solar radiative flux. However, since the average for the two periods was similar it is likely this effect is related to the observed diurnal variations in cooling rather than seasonal differences. The next greatest correlation with DTOS2 was water temperature. Since the average water temperature in June is approximately 2 C warmer than in May this may impact on the cooling potential as the warmer river temperatures will reduce the availability of sensible cooling. For small urban rivers dry periods may result in very shallow levels of flow in the river. This reduction in volume means the water will increase in temperature more rapidly, and at higher water temperatures there
22
E.A. Hathway, S. Sharples / Building and Environment 58 (2012) 14e22
is a reduced capacity for sensible cooling. The River Don is a relatively small river, with a mean flow rate approximately thirteen times smaller than the River Thames as it passes through Kingston [20]. For both these periods the Don’s flow was low: 1.4 m3/s in June and 2.12 m3/s in May. Therefore these results are likely to be representative for smaller urban rivers, but not large water bodies with a more constant yearly temperature. However, further work would be needed to assess this. It is also worth noting that industrial discharges can result in higher water temperatures which will also reduce the capacity for cooling. Even with cooling only being provided in the spring, when the water temperature is cooler, this can have a substantial benefit on comfort. The lack of phased adaptation to the warmer temperatures means spring time heat waves can create greater relative discomfort than in the summer, and so the presence of a water body providing cooling at this time can be highly beneficial. Wind speed may also have implications for different levels of cooling measured at the site since the wind speed was double in June what it was in May. Although higher wind speeds should promote evaporation and, therefore, cooling of the river and potentially the air, this study has shown temperatures at the site more similar to the urban fabric during periods of high wind speeds. Smaller or positive values of DTi do not necessarily represent less cooling from the river, as the higher wind speeds will remove the air from the site more rapidly, and result in greater mixing of the air. Therefore, the correlation between wind speed and positive values of DTi may be due to warmer air being transported to the site. 4.4. Practical implications for the urban form in river corridors The impact of the urban form on the results show the importance of the design of the river corridor, over the simple presence of the water way. The results show that general bioclimatic design principles hold true in the design of river corridors. Greatest cooling is found where there is high amount of tree cover and shading. Any benefits of the river are lost quickly where the site is physically cut off from the water by walls or buildings. Cooling also quickly diminishes when the bankside consists of dark materials such as brick or tarmac with a south west orientation. 5. Conclusions This research presents the results from the study of the microclimate around a small urban river in the UK, representative of many other watercourses in built up areas. Due to the length of the study substantial variations in weather conditions can be considered, alongside the relative impacts of different types of urban typology. The results show significant cooling over the river with an average cooling of nearly 1 C during ambient conditions greater than 20 C. For consecutive hot days in May there was substantial cooling shown over the river of nearly 2 C and 1.5 C cooling on the river bank. This was only shown during the day and was not evident for night time temperature contours across the whole site. In late June, with a similar ambient air temperature, the cooling had diminished. Hourly averages of the two hot periods showed that this cooling tended to be greatest in the mornings with cooling at the river bank varying by 2 C depending on the urban form. Therefore in order to maximise the benefits of this cooling close consideration of the urban design is required. Highly vegetated banks showed much lower temperatures than those banks consisting of only hard
engineering materials. The nature of this surrounding material can have a greater impact on air temperatures than the presence of the river alone. Opening up the streets to the river or the provision of a square gave greater cooling 30 m from the bank than streets shut off from the river. However it was clear that other microclimate factors, such as material albedo and shading, may have been an influence. Therefore, further work into these aspects is required to provide detailed design guidance to gain the most effective cooling from an urban river. Acknowledgements This paper is based on work undertaken as part of the URSULA project funded by the Engineering and Physical Sciences Research Council (grant number EP/F007388/1). The authors are grateful for EPSRC’s support. The views presented in the paper are those of the authors, and cannot be taken as indicative in any way of the position of URSULA colleagues, partners or of EPSRC. All errors are those of the authors alone. References [1] Smith C, Levermore G. Designing urban spaces and buildings to improve sustainability and quality of life in a warmer world. Energ Pol 2008;36: 4558e62. [2] Upmani H, Eliasson I, Lindqvist S. The influence of green areas on nocturnal temperatures in a high latitude city (Göteborg, Sweden). Int J Climatol 1998; 18:681e700. [3] Jansson C. Urban microclimate and surface hydrometeorological processes, thesis. [Stockholm]: Royal Institute of Technology (KTH). 2006. p. 24. [4] Jauregui E. Influence of a large urban park on temperature and convective precipitation in a tropical city. Energy Build 1990;15e16:457e63. [5] Shashua-Bar L, Hoffman ME. Vegetation as a climatic component in the design of an urban street: an empirical model for predicting the cooling effect of urban green areas with trees. Energy Build 2000;31:221e35. [6] Yang F, Lau SSY, Qian F. Summer time heat island intensities in three high rise housing quarters in inner city Shanghai, China. Build Environ 2010;45: 115e34. [7] Taha H. Urban climates and heat islands: albedo, evapotranspiration and anthropogenic heat. Energy Build 1997;25:99e103. [8] Grimmond CSB, Oke TR. An evapotranspiration-interception model for urban areas. Water Resour Res 1991;277:1739e55. [9] Kim YH. Does the restoration of an inner-city stream in Seoul affect local thermal environment. Theor Appl Climatol 2008;92:239e48. [10] Huang l, Zhao D, Wang J, Zhu J, Li J. Scale impacts of land cover and vegetation corridors on urban thermal behavior in Nanjing, China. Theor Appl Climatol 2007;94:241e57. [11] Murakawa S, Sekine T, Narita K. Study of the effects of a river on the thermal environment in an urban area. Energy Build 1990;15:993e1001. [12] Webb BW, Zhang Y. Spatial and seasonal variability in the components of the river heat budget. Hydrol Process 1997;11:79e101. [13] Hannah DM, Malcolm IA, Soulsby C, Youngson AF. Heat exchanges and temperatures within a salmon spawning stream in the Cairngorms, Scotland: seasonal and sub-seasonal dynamics. River Res Appl 2004;20:635e52. [14] CIWEM. Policy position statement on deculverting of watercourses [Internet]. London: Chartered Institute of Water and Environmental Management [updated Dec 2007: cited 2012 Feb 15]. Available from: http://www.ciwem. org/policy-and-international/policy-position-statements; 2007. [15] Lee SE, Sharples S. An analysis of the Urban Heat Island of Sheffield e the impact of a changing climate. In: Kenny P, Brophy V, Lewis JO, editors. Proceedings of the 25th conference on passive and low energy architecture. Dublin: University College Dublin; 2008. p. 396. [16] Barry RG. Atmosphere, weather and climate. 9th ed. London: Routledge; 2009. [17] Pepin N, Bentham D, Taylor K. Modelling lapse rates in maritime uplands of northern England: implications for climate change. Artic Antartic Alpine Res 1999;31:151e64. [18] Rosnow RL, Rosenthal R, Rubin DB. Contrasts and correlations in effect size estimations. Psychol Sci 2000;11:446e53. [19] Hinton P. Statistics explained. 2nd ed. London: Routledge; 2004. [20] CEH. National river flow archive [Internet]. Wallingford (UK): NERC e Centre for Ecology and Hydrology [Updated 2012; cited 2012 Feb 15] Available from: http://www.ceh.ac.uk/data/nrfa/data/search.html; 2010.